In this paper we describe control challenges related to operation of oil wells equipped with Electric Submersible Pumps (ESP) and formalize them in a control problem setting in the language of control system engineers. Then we present a simple dynamic model of an oil well equipped with ESP. This model can be used for controller development. To solve this problem, we propose a Model Predictive Control (MPC) strategy and present experimental results of an MPC controller successfully tested in a large scale test facility with a full scale ESP, live crude oil in an emulated oil well.

No abstract

We present a distributed extremum seeking algorithm for the problem of production optimization of multiple gas lifted wells. The algorithm is based on "synchronization" of production performance gradients for all individual wells. It mimics the manual optimization method employed by production engineers in industry. Thus due to better understanding by industrial specialists, this method may have higher chances of being accepted in the oil and gas industry compared to other data-driven optimization methods. Performance of the proposed algorithm is illustrated by simulations.

Proper allocation and distribution of lift gas is necessary for maximizing total oil production from a eld with gas lifted oil wells. When the supply of the lift gas is limited, the total available gas should be optimally distributed among the oil wells of the eld such that the total production of oil from the eld is maximized. This paper describes a non-linear optimization problem with constraints associated with the optimal distribution of the lift gas. A non-linear objective function is developed using a simple dynamic model of the oil eld where the decision variables represent the lift gas ow rate set points of each oil well of the eld. The lift gas optimization problem is solved using the 'fmincon' solver found in MATLAB. As an alternative and for verication, hill climbing method is utilized for solving the optimization problem. Using both of these methods, it has been shown that after optimization, the total oil production is increased by about 4%. For multiple oil wells sharing lift gas from a common source, a cascade control strategy along with a nonlinear steady state optimizer behaves as a self-optimizing control structure when the total supply of lift gas is assumed to be the only input disturbance present in the process. Simulation results show that repeated optimization performed after the rst time optimization under the presence of the input disturbance has no eect in the total oil production.

No abstract

Process fluctuations are often equivalent to lost production as the necessary margins to process constraints need to take into account the fluctuations. Better operation is achieved by reduced process fluctuations so that the average production may be closer to the constraints. This is achieved by automatic process control which also reduces operator load. Wells which produce oil and free gas (coning) from the reservoir are sensitive to changes in operating conditions, and the gas and oil rates drift over time. This is natural because variations in coned free gas generate variations in wellhead pressure. This will in turn affect the production rate and the well drawdown. In manual operation, a common strategy is to operate with an excess of free gas or use artificial gas lift and manipulate the wellhead choke manually. An automatic wellhead choke control solution is developed to reduce fluctuations and increase production rates. The two main elements in the automatic wellhead choke control are: A robust measure of the controlled variable (i.e. the gas flow rate) A robust control strategy Since the gas flow rate is not directly measured, a gas flow rate estimator was developed, with the premise that no hardware modification should be necessary: The gas fraction in the well stream is determined from the pressure drop from the bottomhole to the wellhead The total multiphase well flow is calculated from either a venturi meter (if available) or the well head choke (based on choke pressure drop, travel, and characteristics) The control strategy is to ensure fast corrections to keep the gas flow rate close to the specified set point, thus enabling a close-to-collapse operation of the established gas cone. The gas flow rate set point is specified to ensure that the well head pressure is sufficiently high, i.e. higher than the pressure downstream the choke. The solution is implemented on several wells at various Statoil operated fields with gas coning into the wells. The solution gives stable well flow rates, and thus enables increased production. The operator workload related to well supervision has also been reduced. The method could also be applied to other that should remain constant, e.g. water cut, water flow, or GOR, as long as the quantity can be estimated (or measured), and will respond to well head choke manipulations.

Ensemble learning in the estimation of flow types and velocities of individual phases in multiphase flow using non-intrusive accelerometers' and process pressure data [Conference presentation]. 2022 IEEE Sensors Applications Symposium (SAS), Sundsvall.

In this paper, we present a water-cut estimator utilizing the function approximation capability of an artificial neural network (ANN). The inputs to the ANN are optical sensor readings in a Red-Eye water-cut meter, which features the near-infrared (NIR) absorption spectroscopy technology. The initial training of the ANNwas done with a data set acquired from our multi-phase flow-loop test facility, which was filled with live oil, water and gas. The test fluid stream was adjusted with good ranges of water-cut and gas-volume fractions which were supposed to cover the situations that can be foreseen in real production. However, clear discrepancies between the outputs of the ANN and the water-cut values from BS&W measurmentswere observedwhen the ANN was applied to actual production data measured by Red-Eye meters installed at two offshore wells. To address this issue and equip the ANN with self-adapting capability in real application, we propose a Bayesian approach to update the parameters of the ANN based on both initial flow-loop data and collected field data. The performance of the adapted ANN on both the data sets shows the effectiveness of the method.